Letter | Published:

A low-energy core-collapse supernova without a hydrogen envelope

Nature volume 459, pages 674677 (04 June 2009) | Download Citation



The final fate of massive stars depends on many factors. Theory suggests that some with initial masses greater than 25 to 30 solar masses end up as Wolf–Rayet stars, which are deficient in hydrogen in their outer layers because of mass loss through strong stellar winds. The most massive of these stars have cores which may form a black hole and theory predicts that the resulting explosion of some of them produces ejecta of low kinetic energy, a faint optical luminosity and a small mass fraction of radioactive nickel1,2,3. An alternative origin for low-energy supernovae is the collapse of the oxygen–neon core of a star of 7–9 solar masses4,5. No weak, hydrogen-deficient, core-collapse supernovae have hitherto been seen. Here we report that SN 2008ha is a faint hydrogen-poor supernova. We propose that other similar events have been observed but have been misclassified as peculiar thermonuclear supernovae (sometimes labelled SN 2002cx-like events6). This discovery could link these faint supernovae to some long-duration γ-ray bursts, because extremely faint, hydrogen-stripped core-collapse supernovae have been proposed to produce such long γ-ray bursts, the afterglows of which do not show evidence of associated supernovae7,8,9.


SN 2008ha was discovered on 7 Nov. 2008 4.1 h ut in the late-type galaxy UGC 12682, at an unfiltered magnitude of 18.8 (ref. 10). A spectrum obtained on 18.18 Nov. 2008 ut (ref. 11) was similar to that of the peculiar supernova SN 2002cx at 10 days past maximum light. The narrow P-Cygni lines are indicative of slowly expanding ejecta, suggesting an outflow velocity 3,000 km s-1 slower than that of SN 2002cx. Our spectra show narrow spectral lines confirming the low velocities of the ejected material (2,300 km s-1) and no hydrogen signature (see Fig. 1 and Supplementary Information).

Figure 1: Spectral evolution of SN 2008ha.
Figure 1

Spectral evolution of SN 2008ha from +8 to +65 days after R-band maximum (which we estimate to have occurred on Julian day JD = 2454787 ± 2). There is no signature of hydrogen features in the spectra, whereas Fe ii, Ti ii, Cr ii, Na iD, O i and Ca ii are clearly detected. The expansion velocity (measured from the minimum of the Fe ii lines) is slowly decreasing from 2,300 to 1,500 km s-1 in the covered spectral range suggesting a very-slow-velocity evolution. The low velocity of SN 2008ha allows us to identify most lines that are normally blended. Fe ii and other metal lines are identified, and those of intermediate-mass elements (Na i, Ca ii, O i) are prominent at all epochs (see Supplementary Fig. 3). After +36 days strong lines of [Ca ii], typical of core-collapse supernovae, become visible, confirming the fast evolution to the nebular phase of SN 2008ha. The low expansion velocity and the fast spectral evolution suggest an extremely low kinetic energy (1–5 × 1049 erg) and ejected mass (0.1–0.5 solar masses) of SN 2008ha. These values are inconsistent with a thermonuclear scenario and quite common in faint core-collapse supernovae.

The spectral evolution is very fast, and [Ca ii] emission at wavelengths λ = 7,291–7,324 Å was detected one month after the explosion (usually it is visible in core-collapse supernovae only after two to three months). The ejecta velocity slowly decreases with time, reaching 1,500 km s-1 in the last spectrum. Such low ejecta velocities (1,000–1,500 km s-1) have been observed in a group of low-luminosity hydrogen-rich core-collapse supernovae12 (faint type IIP supernovae) with extremely narrow P-Cygni spectral lines, but never in thermonuclear supernovae (type Ia supernovae). These objects are believed to be weak explosions (a few times 1050 erg) of massive stars, ejecting only 10-3 solar masses of radioactive material12.

The photospheric spectrum of SN 2008ha is remarkably similar to that of SN 2005cs (see Fig. 2), apart from the absence of hydrogen lines and the strength of the O i λ = 7,774 Å feature, which is more pronounced in the spectrum of SN 2008ha than in that of SN 2005cs. The characteristic type Ia supernova lines (Si ii λ = 6,347–6,371 Å and S ii λ = 5,640 Å), clearly visible in the spectrum of SN 1991T, are distinctly weak or absent in that of SN 2008ha.

Figure 2: Comparison of spectra of SN 2008ha with those of other supernovae.
Figure 2

a, The spectrum of SN 2008ha taken 8 days after maximum is compared with that of the under-luminous type IIP SN 2005cs15 during the hydrogen recombination, with those of the luminous thermonuclear SN 1991T29 (this sub-type of type Ia supernovae has been proposed to share some spectral similarities with SN 2002cx19) and the type Ic SN 2007gr13 at comparable phases. The spectra of SN 2008ha and SN 2005cs are very similar, except for the H lines (always prominent in type IIP supernovae) and the O i feature at λ = 7,774 Å. The photospheric spectra of type Ia and type Ic supernovae both share some similarities with SN 2008ha but only if the spectra are red-shifted (by an ad hoc velocity) or smoothed with a gaussian filter. b, The latest spectrum of SN 2008ha at 65 days past the R-band maximum has not fully completed the transition to the nebular phase. However the remarkably fast evolution of this event means that it can be compared with the nebular spectra of other supernovae. SN 2008ha is very similar to the nebular spectrum of SN 2005cs15 except for the distinct lack of hydrogen features (enlarged in the inset plots). The prominent [Fe ii] lines typical of thermonuclear supernovae and the [O i] lines of stripped-envelope core-collapse supernovae are not apparent in SN 2008ha.

Although at early phases SN 2008ha and SN 2005cs are strikingly similar, we know that photospheric spectra alone can be equivocal in discerning between thermonuclear and core-collapse explosions. The existence of prominent Si ii and S ii lines (combined with a lack of He i and hydrogen features) generally suggests the thermonuclear explosion mechanism. However, Si ii lines are also detected in some stripped-envelope core-collapse supernovae (for example, SN 2007gr13). To complicate matters, the presence of nearby Fe ii, Ti ii and Cr ii lines blended by high photospheric velocities make the identification of weak Si ii rather uncertain and may cause misclassifications.

Additional constraints on the explosion mechanism can be derived from the study of late-time spectra, when the ejecta are optically thin and it becomes easier to probe the nature of the innermost layers. Because of the extremely fast spectral evolution of SN 2008ha, the spectrum at +65 days already shares a remarkable similarity with the nebular spectra of SN 2005cs except for the hydrogen Balmer line (Hα λ = 6,562 Å; Fig. 2b). The near-infrared Ca ii triplet and emission lines due to [Ca ii] λ = 7,291–7,324 Å (strong in core-collapse supernovae and absent in type Ia supernovae) are clearly visible in both objects. The spectrum at +65 days does not show any evidence of the prominent forbidden iron lines which dominate late-time spectra of thermonuclear supernovae. The lack of these features is a strong indication that little 56Ni was synthesized in the explosion, which does not suggest a thermonuclear origin. The [O i] λ = 6,300–6,364 Å feature, which is usually prominent in stripped-envelope core-collapse supernovae, is undetected in the spectra of SN 2005cs and SN 2008ha, probably because of the high ejecta density at these phases.

SN 2008ha is the faintest and lowest-luminosity hydrogen-deficient supernova known. Using a distance modulus of 31.55 mag and a reddening of E(B - V) = 0.076 mag (see Supplementary Information for details), SN 2008ha has a peak magnitude of MR = –14.5 ± 0.3. This is five magnitudes fainter than typical type Ia supernovae, and three magnitudes fainter than the low-luminosity thermonuclear explosions. A ‘pseudo-bolometric’ light curve is shown in Fig. 3a together with light curves of the thermonuclear SN 1991T and other core-collapse supernovae (SN 1998bw14, SN 2007gr and SN 2005cs). SN 2008ha evolves much more rapidly than these other supernovae and has a maximum luminosity comparable with those of low-luminosity type IIP supernovae12,15.

Figure 3: Pseudo-bolometric and absolute R-band light curves of SN 2008ha.
Figure 3

a, We compare a pseudo-bolometric light curve of SN 2008ha with those of other supernova types (error bars show the standard errors). The pseudo-bolometric light curve of SN 2008ha was computed, using our R-band observations and SN 2005hk as references (after having time-stretched the data of SN 2005hk to fit the time evolution of SN 2008ha). The light curve of SN 2008ha is faster than those of most other supernovae. Considering the low expansion velocity, the light curve of SN 2008ha is inconsistent with a thermonuclear explosion of 1.4 solar masses, typical of type Ia supernovae. The peak luminosity is similar to SN 2005cs, suggesting possibly a stripped-core analogue of this type of explosion15, or the weak explosion of a very massive star with black-hole formation1,2,3 b, The R-band light curve of SN 2008ha is shown together with those of SN 1998bw14 and the afterglows of two long-duration gamma-ray bursts (error bars show the standard errors). The detection of ref. 10 and magnitudes from the Bright Supernova website (http://www.rochesterastronomy.org/snimages/) are shown as triangles. GRB 060614 showed no evidence for a supernova signature in its R-band afterglow light curve, leading the authors of refs 7, 8 and 9 to suggest that a faint supernova (fainter than -13.7 at peak) would be required to be consistent with the canonical physical production mechanism for long gamma-ray bursts. The horizontal line (blue dotted line) is the host galaxy of GRB 060614. The light curve of SN 2008ha shows, for the first time, that such faint hydrogen poor core-collapse supernovae (even though SN 2008ha was slightly brighter than -13.7 at maximum (being R = -14.5 ± 0.3), do exist. As a comparison, the afterglow of a bright gamma-ray burst (GRB 030329) consistent with the explosion of a SN 1998bw-like event is also shown30. In that case the flux excess with respect to the afterglow (green dotted line) was partially due to the SN 1998bw event.

The rapidly evolving light curve of SN 2008ha, together with the modest ejecta velocities, implies very low kinetic energy and ejected mass. We roughly estimate these quantities using a toy model based on Arnett’s equations16 (see Supplementary Information), obtaining an ejecta mass of 0.1–0.5 solar masses and a kinetic energy in the range 1–5 × 1049 erg. The ejected mass is significantly smaller than the canonical 1.4 solar masses expected for thermonuclear supernovae (ref. 17) (with the caveat that the physical values can be better constrained with more accurate modelling). The kinetic energy is also smaller than suggested by models of the pure deflagration scenario, which were proposed to explain SN 2002cx-like events. We also estimate the mass of 56Ni produced in the explosion of SN 2008ha to be 0.003–0.005 solar masses (see Supplementary Information). This amount is very close to that observed in sub-luminous type IIP supernovae12,15, but is two orders of magnitude smaller than in normal type Ia supernovae (0.4–0.8 solar masses of 56Ni , ref. 18). We cannot exclude that some exotic thermonuclear explosion might be consistent with the observed low energy and fast light curve evolution of SN 2008ha, but the observational comparisons indicate that it is more likely that SN 2008ha was produced in the low-energy core-collapse explosion of a hydrogen-deficient massive star.

This discovery has important implications for the origin of some gamma-ray bursts. Several nearby long-duration gamma-ray bursts show evidence of an accompanying bright, highly energetic, envelope-stripped core-collapse supernova in their light curves and spectra. Two long gamma-ray bursts (GRB 060614 and GRB 060505, refs 7–9) were close enough that the presence of an associated supernova could be excluded down to limiting absolute magnitudes of MR ≈ -12.3 to -13.7. One possible explanation is that they were accompanied by an extremely sub-luminous, hydrogen-poor core-collapse supernova. The discovery of SN 2008ha is the first evidence that such supernovae do exist (see Fig. 3b).

The observed properties of SN 2008ha are undeniably similar to the group of supernovae similar to SN 2002cx6,19,20,21,22, and it is perhaps the most extreme object of its kind (see Supplementary Information for more details on this supernova group). If SN 2008ha is more plausibly explained by core collapse then, by implication, all SN 2002cx events could possibly be explosions of this nature. So far they have been interpreted as pure thermonuclear deflagrations of 1.4-solar-mass (Chandrasekhar mass) white dwarfs23, although their observed characteristics deviate significantly from those of type Ia supernovae. Their intrinsic faintness and broad light curves are at odds with the luminosity versus light curve shape relation24 which characterizes type Ia supernovae and itself is a consequence of comparable ejecta masses (ref. 17). The spectra of SN 2002cx-like supernovae are quite bizarre: before maximum they show similarities to luminous type Ia supernovae19 (for example, SN 1991T), after maximum they are very similar to those of SN 2008ha (see Fig. 4a) and at late time the spectra resemble those of faint core-collapse supernovae. A comparison of an unpublished late-time spectrum of SN 2005hk21,22 (the best-studied SN 2002cx-like event), with that of the sub-luminous type IIP SN 1997D25 is shown in Fig. 4b. The lack of forbidden lines of oxygen and iron in both spectra, and the presence of P-Cygni-type lines of Fe ii about 400 days after explosion (in the case of SN 2005hk) is evidence of high density (108 cm-3, ref. 22). The authors of ref. 22, indeed, attempted to model a spectrum of SN 2005hk at 228 days by combining a photospheric spectrum and a nebular spectrum. However, the model required to reproduce the photospheric phase (a W7 model26 scaled down to an energy of 3 × 1050 erg) was unable to reproduce the high density of the inner ejecta at late phases. High density and low energy are sometimes found in core-collapse supernovae, particularly in faint type IIP supernovae12,15.

Figure 4: Spectra of SN 2008ha and SN 2002cx-like events.
Figure 4

a, The spectrum of SN 2008ha is compared with those of SN 2002cx6,19,20 and SN 2005hk21,22 at 2–3 weeks after maximum light. The most significant differences between the spectra are due to a different degree of line blending. The spectra of supernovae 2005hk and 2002cx are shifted by -3,000 km s-1 to align the absorptions. b, The nebular spectrum of SN 2005hk at 400 days is compared with that of the faint hydrogen-rich supernova SN 1997D25. In analogy to SN 1997D, the nebular spectrum of SN 2005hk is rich in permitted Fe ii lines, clear evidence that the ejecta are still dense and not completely transparent. The [Fe ii] and [Fe iiI] lines typical of thermonuclear supernovae, and the [O i] typical of stripped-envelope core-collapse supernovae, are not detected. In SN 1997D the [O i] feature appeared only about six months later25. The spectrum of SN 2005hk was taken on 27 November 2006 at the Very Large Telescope VLT+FORS2 (programme ID 078.D0246) (Stanishev, V. et al., manuscript in preparation).

Additional support comes from the detection of He i lines in SN 2007J27, a supernova showing remarkable similarities with SN 2002cx at an earlier phase28. Helium lines have never been detected in thermonuclear supernovae. There is quite significant evidence for SN 2008ha being a core-collapse supernova, and this family of objects could plausibly be of the same nature. Future observations of similar events will help us to understand whether they are a form of thermonuclear explosions, low-luminosity core-collapse supernovae from stripped stars of moderate mass, or the deaths of very massive stars inducing black-hole formation and fall-back.


  1. 1.

    , & Supernovae, jets, and collapsars. Astrophys. J. 550, 410–425 (2001)

  2. 2.

    et al. How massive single stars end their life. Astrophys. J. 591, 288–300 (2003)

  3. 3.

    , & Explosive nucleosynthesis from gamma-ray burst and hypernova progenitors: direct collapse versus fallback. Astrophys. J. 650, 1028–1047 (2006)

  4. 4.

    Evolution of 8–10 solar mass stars toward electron capture supernovae. I. Formation of electron-degenerate O + Ne + Mg cores. Astrophys. J. 277, 791–805 (1984)

  5. 5.

    , & Explosions of O-Ne-Mg cores, the Crab supernova, and subluminous type II-P supernovae. Astron. Astrophys. 450, 345–350 (2006)

  6. 6.

    et al. SN 2002cx: The most peculiar known type Ia supernova. Publ. Astron. Soc. Pacif. 115, 453–473 (2003)

  7. 7.

    et al. An enigmatic long-lasting γ-ray burst not accompanied by a bright supernova. Nature 444, 1050–1052 (2006)

  8. 8.

    et al. A novel explosive process is required for the γ-ray burst GRB 060614. Nature 444, 1053–1055 (2006)

  9. 9.

    et al. No supernovae associated with two long-duration γ-ray bursts. Nature 444, 1047–1049 (2006)

  10. 10.

    , , & Supernova 2008ha in UGC 12682. Central Bureau Electr. Telegr. 1567, (2008)

  11. 11.

    et al. Supernova 2008ha in UGC 12682. Central Bureau Electr. Telegr. 1576, (2008)

  12. 12.

    et al. Low-luminosity type II supernovae: spectroscopic and photometric evolution. Mon. Not. R. Astron. Soc. 347, 74–94 (2004)

  13. 13.

    et al. The carbon-rich type Ic SN 2007gr: the photospheric phase. Astrophys. J. 673, L155–L158 (2008)

  14. 14.

    et al. The metamorphosis of SN 1998bw. Astrophys. J. 555, 900–917 (2001)

  15. 15.

    et al. SN 2005cs in M51. II. Complete evolution in the optical and the near-infrared. Mon. Not. R. Astron. Soc. 394, 2266–2282 (2009)

  16. 16.

    Type I supernovae. I—Analytic solutions for the early part of the light curve. Astrophys. J. 253, 785–797 (1982)

  17. 17.

    , , & A common explosion mechanism for type Ia supernovae. Science 315, 825–828 (2007)

  18. 18.

    et al. Can differences in the nickel abundance in Chandrasekhar-mass models explain the relation between the brightness and decline rate of normal type IA supernovae? Astrophys. J. 547, 988–994 (2001)

  19. 19.

    et al. Reading the spectra of the most peculiar type Ia supernova 2002cx. Publ. Astron. Soc. Pacif. 116, 903–908 (2004)

  20. 20.

    et al. Late-time spectroscopy of SN 2002cx: the prototype of a new subclass of type Ia supernovae. Astron. J. 132, 189–196 (2006)

  21. 21.

    et al. The peculiar SN 2005hk: do some type Ia supernovae explode as deflagrations? Publ. Astron. Soc. Pacif. 119, 360–387 (2007)

  22. 22.

    et al. The evolution of the peculiar type Ia supernova SN 2005hk over 400 days. Astrophys. J. 680, 580–592 (2008)

  23. 23.

    , & Deflagrations and detonations in thermonuclear supernovae. Phys. Rev. Lett. 92, 211102 (2004)

  24. 24.

    The absolute magnitudes of type IA supernovae. Astrophys. J. 413, L105–L108 (1993)

  25. 25.

    et al. The fading of supernova 1997D. Mon. Not. R. Astron. Soc. 322, 361–368 (2001)

  26. 26.

    , & Accreting white dwarf models of type I supernovae. III—Carbon deflagration supernovae. Astrophys. J. 286, 644–658 (1984)

  27. 27.

    et al. Supernova 2007J in UGC 1778. Central Bureau Electr. Telegr. 926, 1 (2007)

  28. 28.

    et al. Supernova 2007J in UGC 1778. Central Bureau Electr. Telegr. 817, 1 (2007)

  29. 29.

    et al. The peculiar type IA SN 1991T—Detonation of a white dwarf? Astrophys. J. 384, L15–L18 (1992)

  30. 30.

    et al. Photometry and spectroscopy of GRB 030329 and its associated supernova 2003dh: the first two months. Astrophys. J. 599, 394–407 (2003)

Download references


This work, conducted as part of the European Science Foundation EURYI Awards scheme, was supported by funds from the Participating Organisations of EURYI and the EC Sixth Framework Programme. The work of S.B., E.C. and M.T. was supported by grants of the PRIN of Italian Ministry of University and Science Research. This paper is based on observations collected at TNG, NOT, LT (La Palma Canary Island, Spain), at Ekar (Asiago Observatory, Italy), at the Begues Observatory and Arguines Observatory telescopes (Barcelona and Segorbe, Spain), at the Taurus Hill Observatory (Varkaus, Finland), at the Calar Alto Observatory (Spain) and at the ESO-UT2 (Paranal, Chile). Our analysis included data from the SUSPECT Archive (http://bruford.nhn.ou.edu/~suspect/index1.html). This manuscript made use of information contained in the Bright Supernova web pages (D. Bishop), as part of the Rochester Academy of Sciences.

Author information


  1. Astrophysics Research Centre, School of Mathematics and Physics, Queen’s University Belfast, Belfast BT7 1NN, UK

    • S. Valenti
    • , A. Pastorello
    •  & S. J. Smartt
  2. INAF Osservatorio Astronomico di Padova, Vicolo dell' Osservatorio 5, I-35122 Padova, Italy

    • E. Cappellaro
    • , S. Benetti
    • , P. A. Mazzali
    • , A. Harutyunyan
    •  & L. Zampieri
  3. Max-Planck-Institut für Astrophysik, Karl-Schwarzschild-Strasse 1, D-85741 Garching bei München, Germany

    • P. A. Mazzali
    •  & S. Taubenberger
  4. Begues Observatory, Santpere 6 Casa 22, 08859 Begues, Barcelona, Spain

    • J. Manteca
  5. Spitzer Science Center, California Institute of Technology, 1200 E. California Blvd, Pasadena, California 91125, USA

    • N. Elias-Rosa
  6. Calle de la Guardia Civil 22, 46020 Valencia, Spain

    • R. Ferrando
  7. Fundación Galileo Galilei-INAF, Telescopio Nazionale Galileo, E-38700 Santa Cruz de la Palma, Tenerife, Spain

    • A. Harutyunyan
  8. Taurus Hill Observatory, Härcämäentie 88, 79480 Kangaslampi, Finland

    • V. P. Hentunen
    •  & M. Nissinen
  9. Tuorla Observatory, Department of Physics and Astronomy, University of Turku, Väisäläntie 20, FI-21500 Piikkiö, Finland

    • V. P. Hentunen
  10. INAF Osservatorio Astronomico di Trieste, Via Tiepolo 11, I-34131 Trieste, Italy

    • E. Pian
  11. INAF Osservatorio Astronomico di Catania, 78 Via S. Sofia, 95123 Catania, Italy

    • M. Turatto


  1. Search for S. Valenti in:

  2. Search for A. Pastorello in:

  3. Search for E. Cappellaro in:

  4. Search for S. Benetti in:

  5. Search for P. A. Mazzali in:

  6. Search for J. Manteca in:

  7. Search for S. Taubenberger in:

  8. Search for N. Elias-Rosa in:

  9. Search for R. Ferrando in:

  10. Search for A. Harutyunyan in:

  11. Search for V. P. Hentunen in:

  12. Search for M. Nissinen in:

  13. Search for E. Pian in:

  14. Search for M. Turatto in:

  15. Search for L. Zampieri in:

  16. Search for S. J. Smartt in:

Corresponding author

Correspondence to S. Valenti.

Supplementary information

PDF files

  1. 1.

    Supplementary Information

    This file contains Supplementary Methods and Data, Supplementary Tables 1-2, Supplementary Figures 1-4 with Legends and Supplementary References.

About this article

Publication history






Further reading


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.